Social network analysis and mining refers to a set of methods and techniques to analyze such networks and discover useful patterns. It covers a large range of topics such as network evolution, community detection, information diffusion, influence computation, link prediction and recommendation, security management, and anomaly detection. Most of the existing methods exploit mainly one mode data such as friendship links, or two mode data like actor-by-event attendance. They also assume the existence of only one kind of links in a given network. However, there exist heterogeneous information networks that have different types of nodes and links among nodes. For instance, node types can be researchers, topics and publications while "write" is a link type from researcher to publication. Moreover, a network may be multi-modal (multi-relational) to express for example the fact that a researcher attends a conference with given roles. The objective of this talk is to overview recent studies on mining and analyzing heterogeneous information networks (HINs) with a focus on influential actor identification, community detection, and link prediction. We also present our current work towards HIN mining where formal concept analysis (with variants) is exploited to detect associations and clusters from such networks.

Short bio: Rokia Missaoui received her Ph.D. in Computer Science (CS) in 1988 from Université de Montréal, Canada. She is a Full Professor in the Department of CS and Engineering at Université du Québec in Outaouais (UQO). Before joining UQO, she was a professor at UQAM (Université du Québec à Montréal) between 1987 and 2002. Currently, she is the head of the LARIM laboratory. Her research interests include data mining and warehousing, formal concept analysis, and social network analysis.